CN109377429A - A kind of recognition of face quality-oriented education wisdom evaluation system - Google Patents
A kind of recognition of face quality-oriented education wisdom evaluation system Download PDFInfo
- Publication number
- CN109377429A CN109377429A CN201811345573.2A CN201811345573A CN109377429A CN 109377429 A CN109377429 A CN 109377429A CN 201811345573 A CN201811345573 A CN 201811345573A CN 109377429 A CN109377429 A CN 109377429A
- Authority
- CN
- China
- Prior art keywords
- face
- recognition
- image
- evaluation
- wisdom
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 66
- 230000001815 facial effect Effects 0.000 claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 25
- 239000013598 vector Substances 0.000 claims abstract description 24
- 239000011521 glass Substances 0.000 claims abstract description 20
- 238000005286 illumination Methods 0.000 claims abstract description 18
- 238000013135 deep learning Methods 0.000 claims abstract description 15
- 238000001514 detection method Methods 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 230000008439 repair process Effects 0.000 claims description 3
- 230000005484 gravity Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 9
- 230000008859 change Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000011161 development Methods 0.000 description 4
- 210000000887 face Anatomy 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 210000003128 head Anatomy 0.000 description 3
- 238000012360 testing method Methods 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 206010044074 Torticollis Diseases 0.000 description 1
- 238000012550 audit Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000002996 emotional effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 208000018197 inherited torticollis Diseases 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000001144 postural effect Effects 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
- G06Q50/205—Education administration or guidance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Educational Technology (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of recognition of face quality-oriented education wisdom evaluation methods characterized by comprising to Face datection, and to facial modeling；Facial image is pre-processed, face standard drawing is generated；Face standard drawing is generated into the human face recognition model based on deep learning, calculates the cosine similarity obtained in the face feature vector based on human face recognition model and database between existing face database vector；Result judgement face recognition result based on cosine similarity；If recognition of face success enters overall evaluation system and carries out overall merit, and generates overall merit handbook.Recognition of face quality-oriented education wisdom evaluation method of the invention, face recognition technology is applied in quality-oriented education wisdom evaluation method, and propose the method that can effectively eliminate illumination and glasses shelter, it is also proposed that unrelated with illumination change, not by the face identification method of attitudes vibration.
Description
Technical field
The present invention relates to face recognition application technical fields, evaluate more particularly to a kind of recognition of face quality-oriented education wisdom
System.
Background technique
Face recognition technology rapidly develops in the past few years, currently, face recognition technology is actually raw in outdoor environment etc.
It cannot cope in environment living, usually only use indoors very well.The difficult point of recognition of face remain illumination change, postural change,
Change of age is blocked, these have an impact face recognition algorithms used by face identification system.
Currently, mainly carrying out Face datection using haar feature on the market, have illumination effect is larger, detection speed is slow,
Face is aligned slow-footed problem.Using PCA recognition of face, when image dimension is very big, recognition speed is very slow, solves way
Diameter is dimensionality reduction to be carried out to image, but dimensionality reduction will lead to and lose a large amount of details, while being illuminated by the light and being affected, but illumination condition is not
Meanwhile comparison result will have a greatly reduced quality.
Face identification method based on deep learning, deep neural network do not take artificial extracting mode, this saves on
It is artificial to extract the spent plenty of time, the intelligent recognition process of face face is completed with prestissimo.Deep neural network
Have huge spread with traditional network: first point, be the stepped construction of modular；Second point, when adjustment neural network weight
When, weight will be automatically close to optimum point.It therefore, there is no need to the training and supervision by early period, finally obtain a perfection
Data.
Most of studies have shown that takes the face identification system of artificial neuron method, whether in robustness, fault-tolerant
Property still identify that all there is in terms of accuracy stronger advantage.
In real life environments, facial image by illumination, posture, at the age, the variations such as block and cause recognition of face difficulty.
Therefore, in real life environments, face recognition technology cannot be satisfactory, in recent years, has carried out many researchs in this field,
There is very big progress, but still reaches to less than satisfied effect.
In addition, some wisdom evaluation systems are had increasing need for the management of student with the propulsion of quality-oriented education, and it is current
There are no occur carrying out the recognition of face quality-oriented education wisdom evaluation system of identification and evaluation using face recognition technology.
Summary of the invention
Based on this, the object of the present invention is to provide a kind of recognition of face quality-oriented education wisdom evaluation systems.
A kind of recognition of face quality-oriented education wisdom evaluation method, comprising:
To Face datection, and to facial modeling；
Facial image is pre-processed, face standard drawing is generated；
Face standard drawing is generated into the human face recognition model based on deep learning, calculates and obtains based on human face recognition model
Cosine similarity in face feature vector and database between existing face database vector；
Result judgement face recognition result based on cosine similarity；
If recognition of face success enters overall evaluation system and carries out overall merit, and generates overall merit handbook.
It is further preferred that described to Face datection, and to facial modeling, comprising: based on LBP feature and
AdaBoost carries out Face datection, and is positioned based on points distribution models algorithm to 68 human face characteristic points；
It is described that Face datection is carried out based on LBP feature and AdaBoost, including cromogram is converted into grayscale image, any
In neighborhood, using centre of neighbourhood pixel as threshold value, the gray value of 8 adjacent pixels is compared with it, if neighborhood territory pixel value
Greater than center pixel value, then the position of the pixel is marked as 1, is otherwise 0；
8 pixels in neighborhood, which are compared, generates 8 bits, and 8 bit is converted to decimal number,
The LBP value of the centre of neighbourhood pixel is calculated.
It is further preferred that the LBP value that the centre of neighbourhood pixel is calculated, comprising:
If (xc,yc) be center pixel coordinate, p be neighborhood p-th of pixel, ipFor the gray value of neighborhood territory pixel, icFor
The gray value of center pixel, LBP (xc,yc) be center pixel LBP value；X is neighborhood territory pixel value-center pixel value difference, s
It (x) is sign function；
Then
Further, the LBP value of the centre of neighbourhood pixel is calculated,
It is further preferred that described carry out Face datection based on LBP feature and AdaBoost, further includes: by the LBP
(xc,yc) it is sent into AdaBoost classifier, and classify.
It is further preferred that the AdaBoost classifier includes multiple cascade classifiers；
It is described by the LBP (xc,yc) it is sent into AdaBoost classifier, and classify, comprising: the AdaBoost points
Class device is detected using multiple dimensioned sliding window, the window that each scale interception size is 20*20, and window is put into multiple cascades point
Judgement is face in class device, if face, then the window is by all cascade classifiers, if not face, then should
Window is excluded in a certain cascade classifier.
It is further preferred that described includes: using face earth axes building face 3D mark to facial image pretreatment
Quasi-mode type estimates human face posture, carries out face image correcting, cuts and be aligned, and carries out illumination pretreatment to facial image, eliminates
Influence of the illumination to face.
It is further preferred that described pre-process facial image further include: carry out Glasses detection to facial image, extract eye
Mirror image obtains denoising image.
It is further preferred that described carry out Glasses detection to facial image, glasses image is extractd, obtains denoising image, packet
It includes: if detecting including glasses image, carrying out binary conversion treatment using maximum variance between clusters, obtain bianry image, using opening
The not connected smaller image border in part is eliminated in operation, and isolated marginal point, removal weight are then removed using the method for connected domain
The heart is in the edge below of picture altitude 1/2, connects remaining edge, then transversal scanning using the method for 2 iteration closed operations
Image, the edge by edge breaks length lower than 1/6 picture traverse are attached, and form rims of spectacle mask image, complete eye
The positioning of mirror frame, finally according to mask image and the glasses periphery colour of skin, repairs frame using linear interpolation mode, obtains
Denoise image.
It is further preferred that described generate the human face recognition model based on deep learning for face standard drawing, calculates and obtain
Cosine similarity in feature vector and database based on human face recognition model between existing face database vector, comprising: logical
Deep learning training human face recognition model is crossed, face characteristic is extracted to face standard drawing based on human face recognition model, to extraction
Face feature vector carries out PCA dimensionality reduction, calculates the distance between the human face characteristic point of face feature vector representative.
It is further preferred that the overall evaluation system includes all evaluation units, moon evaluation unit, term evaluation unit,
And comprehensive score result unit, Suggestions for Development unit, personal integral unit, comparation and assessment unit；
The overall merit include academic record evaluation, morality evaluation, body and mind Qualities Evaluation, affective state evaluation,
Campus performance evaluation.
Compared with the existing technology, recognition of face quality-oriented education wisdom evaluation method of the invention, face recognition technology is answered
For in quality-oriented education wisdom evaluation method, and the method that can effectively eliminate illumination and glasses shelter is proposed, also
Propose it is unrelated with illumination change, not by the face identification method of attitudes vibration.
In order to better understand and implement, the invention will now be described in detail with reference to the accompanying drawings.
Detailed description of the invention
Fig. 1 is the flow diagram of recognition of face quality-oriented education wisdom evaluation method of the invention.
Fig. 2 is the schematic diagram of multiple dimensioned sliding window detection.
Fig. 3 is sliding window scanning schematic diagram.
Fig. 4 is the search schematic diagram of human face characteristic point.
Specific embodiment
Embodiment described in following exemplary embodiment does not represent all embodiment party consistent with this disclosure
Formula.On the contrary, they are only the device and side consistent with some aspects as detailed in the attached claim, the disclosure
The example of method.
It is only to be not intended to be limiting the disclosure merely for for the purpose of describing particular embodiments in the term that the disclosure uses.
The "an" of the singular used in disclosure and the accompanying claims book, " described " and "the" are also intended to including majority
Form, unless the context clearly indicates other meaning.It is also understood that term "and/or" used herein refers to and wraps
It may be combined containing one or more associated any or all of project listed.
Referring to Fig. 1, Fig. 1 is the flow diagram of recognition of face quality-oriented education wisdom evaluation method of the invention.The present invention
Recognition of face quality-oriented education wisdom evaluation method, include the following steps.
In a step 101, to Face datection, and to facial modeling.
In a step 102, facial image is pre-processed, generates face standard drawing.
In step 103, face standard drawing is generated into the human face recognition model based on deep learning, calculates and obtains based on people
Cosine similarity in the face feature vector and database of face identification model between existing face database vector.
At step 104, the result judgement face recognition result based on cosine similarity.
In step 105, if recognition of face success, enter overall evaluation system and carry out overall merit, and generate synthesis
Evaluation handbook.
In the present embodiment, in a step 101, described to Face datection, and to facial modeling, comprising: it is based on
LBP feature and AdaBoost carry out Face datection, and are positioned based on points distribution models algorithm to 68 human face characteristic points, this
68 human face characteristic points include eyes, nose, mouth, eyebrow and chin etc..
It is described that Face datection is carried out based on LBP feature and AdaBoost, including cromogram is converted into grayscale image, any
In neighborhood, using centre of neighbourhood pixel as threshold value, the gray value of 8 adjacent pixels is compared with it, if neighborhood territory pixel value
Greater than center pixel value, then the position of the pixel is marked as 1, is otherwise 0.
8 pixels in neighborhood, which are compared, generates 8 bits, and 8 bit is converted to decimal number,
The LBP value of the centre of neighbourhood pixel is calculated.
Multi-pose Face can quickly be detected, together to illumination-insensitive using LBP feature and AdaBoost classifier
When standard normalization is carried out to face, eliminate influence of the illumination to face, prevent face from turning blue, is rubescent, by shelter
Removal operation can effectively prevent illumination and shelter to influence.
In above-mentioned realization, the LBP value that the centre of neighbourhood pixel is calculated, comprising:
If (xc,yc) be center pixel coordinate, p be neighborhood p-th of pixel, ipFor the gray value of neighborhood territory pixel, icFor
The gray value of center pixel, LBP (xc,yc) be center pixel LBP value；X is neighborhood territory pixel value-center pixel value difference, s
It (x) is sign function；
Then
Further, the LBP value of the centre of neighbourhood pixel is calculated,
It is described that Face datection is carried out based on LBP feature and AdaBoost, further includes: by the LBP (xc,yc) be sent into
AdaBoost classifier, and classify.
In foregoing description, the AdaBoost classifier includes multiple cascade classifiers.
It is described by the LBP (xc,yc) it is sent into AdaBoost classifier, and classify, comprising: the AdaBoost points
Class device is detected using multiple dimensioned sliding window, the window that each scale interception size is 20*20, and window is put into multiple cascades point
Judgement is face in class device, if face, then the window is by all cascade classifiers, if not face, then should
Window is excluded in a certain cascade classifier.
Referring to Fig. 2, Fig. 2 is the schematic diagram of multiple dimensioned sliding window detection.Multiple dimensioned sliding window in foregoing description, which detects, includes
Multiple dimensioned search is to reduce to image by certain ScaleFactor, every to reduce once, on the image after diminution
Carry out the Face datection that size is 20*20.
And single scale search is in the image for narrowing down to a certain scale, on x, two dimensions of y (such as by a fixed step size
The window scanning for 2pixel) carrying out 20*20, the window of interception is sent in cascade classifier and carries out feature extraction and judgement.
Referring to Fig. 3, Fig. 3 is sliding window scanning schematic diagram.As can be seen that the number of windows to be judged of single scale sliding window operation
About (height-20)/ystep* (width-20)/xstep, by taking image to be detected of 100*100 as an example, xstep, ystep
In the case where 2, the window number for needing to judge is about 1600；In multiple dimensioned situation, each scale will carry out sliding window, so
Total detection window number quantity can be much larger.Search space is all to be faced in terms of efficiency based on multiple dimensioned sliding window detection algorithm greatly
Main problem.
Face location is got through the above steps, is passed to the good points distribution models of pre-training, is each human face characteristic point
Construct local feature.
Referring to Fig. 4, Fig. 4 is the search schematic diagram of human face characteristic point.Local feature be used near human face characteristic point into
Row search (can search in the rectangle frame near human face characteristic point, can also search for along normal direction), seeks in an iterative manner
Look for new human face characteristic point matching position.To prevent illumination variation, local feature generally uses Gradient Features to describe.
In a step 102, described includes: to construct face 3D standard using face earth axes to facial image pretreatment
Model estimates human face posture, carries out face image correcting, cuts and be aligned, and carries out illumination pretreatment to facial image, eliminates light
According to the influence to face.
By 68 human face characteristic points of positioning, world coordinate system rotation, transition matrix become 3D point from world coordinate system
It changes in camera coordinates system, that is, world coordinate system (3D), 2Dlandmark input image, camera seat is completed by algorithm
Mapping and Converting and calibration between mark system.
Face 3D model is constructed using the parameter of existing input world coordinate system, passes through 7 human face characteristic points of face
Coordinate is projected, and the posture of face is estimated.
Final result is three Eulerian angles: Yaw: shaking the head, left positive right negative；Pitch: nodding, it is upper it is negative under just；Roll: yaw
(torticollis), the negative right side in a left side is just.Affine transformation, correction are carried out to facial image according to Eulerian angles, keeps face in an intermediate position, makes one
Face is transposed to standard faces.
It is described that facial image is pre-processed further include: Glasses detection is carried out to facial image, glasses image is extractd, is gone
It makes an uproar image.
It is described that Glasses detection is carried out to facial image in foregoing description, glasses image is extractd, denoising image is obtained, comprising:
If detecting including glasses image, binary conversion treatment is carried out using maximum variance between clusters, bianry image is obtained, using opening operation
The not connected smaller image border in part is eliminated, isolated marginal point is then removed using the method for connected domain, removes center of gravity
In the edge below of picture altitude 1/2, remaining edge is connected using the method for 2 iteration closed operations, then transversal scanning image,
Edge by edge breaks length lower than 1/6 picture traverse is attached, and forms rims of spectacle mask image, completes rims of spectacle
Positioning, finally according to mask image and the glasses periphery colour of skin, repairs frame using linear interpolation mode, obtains denoising figure
Picture.
In step 103, described that face standard drawing is generated into the human face recognition model based on deep learning, it calculates and obtains base
Cosine similarity in the feature vector of human face recognition model and database between existing face database vector, comprising: pass through
Deep learning trains human face recognition model, face characteristic is extracted to face standard drawing based on human face recognition model, to the people of extraction
Face feature vector carries out PCA dimensionality reduction, calculates the distance between the human face characteristic point of face feature vector representative.
Accelerated based on the human face recognition model of mass data training using GPU using the face identification method of deep learning
Operation greatly improves face identification rate and recognition rate.
In foregoing description, using VGG16 layers of convolutional neural networks, the input of network is the image of 224*224*3 size, defeated
It is image classification result out.
By acquiring human face photo, database and label file are established.The script for generating lmdb format is established, mean value is generated
Script file.Firstly, database is taken VGG11 training network, the caffemodel obtained after VGG11 training network is sent
Fintune is carried out to VGG16.
Model file is generated after duplicate repetitive exercise, standard faces image is extracted into VGG16 according to model file
Full articulamentum feature vector.
Standard faces image is inputted into VGG, extracts the value of the full articulamentum of VGG as face feature vector, due to fc7's
4096 dimensional vectors are excessive, then carry out dimensionality reduction using PCA, are down to 500 dimensions.
By comparing the distance between feature vector, the gap between two different faces can be measured.
In step 105, the overall evaluation system includes all evaluation units, moon evaluation unit, term evaluation unit, with
And comprehensive score result unit, Suggestions for Development unit, personal integral unit, comparation and assessment unit.
The overall merit include academic record evaluation, morality evaluation, body and mind Qualities Evaluation, affective state evaluation,
Campus performance evaluation.
In foregoing description, after recognition of face success, teacher can various performances according to each student in school and classroom,
By the face of mobile phone candid photograph student, the relevant informations such as the performance of input student at that time is given a mark or evaluated.Overall evaluation system
Information will be automatically recorded in real time, is pressed weekly according to the requirement of school, the moon, carries out overall merit in term, formed to each
The development processes such as morality, academic record, body and mind quality, emotional attitude, the campus performance of student and situation carry out big data
Analysis and value judgement, automatically generate the overall merit handbook of each student.
The practical application of recognition of face quality-oriented education wisdom evaluation method of the invention can be with are as follows:
Log in school backstage, typing student information.
The evaluable quality-oriented education field of typing.
Log in the total backstage of platform, the quality-oriented education field that audit school submits.
The APP for being mounted with overall evaluation system is opened, login account password is inputted.
Recording of growing up subfield is selected in homepage, evaluation is clicked, into the recognition of face page.
The face that mobile phone camera alignment will be known others, is identified；Or grade's button is clicked, selection grade,
Class is not searched.
After recognition of face succeeds or finds student, clicks and the student's head portrait to score is needed to enter the scoring page, commented
Point.
Selection needs the column that scores to score, each is minimum to comment 1 point, and highest can comment 5 points.
It selects to exit if entering the scoring page without scoring, the students' union identified is saved in be evaluated
List；The evaluation page can be entered by clicking list, and long-pressing list can be removed to be evaluated；The student evaluated can look into from honor roll
See class and school's ranking list.
Clicking student's head portrait can carry out checking scoring details；Scoring details can check the comprehensive score of the student, development
It is recommended that and personal integral statistics.
Clicking comparison column can check that the personal scoring with class, individual and age, individual and school compares.
The medal that the student obtains can be checked by clicking personal medal；Medal wall can check all medals and above and below each academic year
Term medal obtained.
Recognition of face quality-oriented education wisdom evaluation method of the invention constructs the perfect people based on deep learning
Face identifies system and overall evaluation system, and recognition of face system includes three basic structures: the pretreatments of data, deep learning with
And identification structure.The process of recognition of face mainly includes two aspects: the process of trained process, test.Trained process needs
A trained model is established, data obtained in training are used in the sample of test, the structural key of deep learning
Component part be module stacking, the use of module number must be chosen according to experiment effect.
Compared with the existing technology, recognition of face quality-oriented education wisdom evaluation method of the invention, face recognition technology is answered
For in quality-oriented education wisdom evaluation method, and the method that can effectively eliminate illumination and glasses shelter is proposed, also
Propose it is unrelated with illumination change, not by the face identification method of attitudes vibration.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to protection of the invention
Range.
Claims (10)
1. a kind of recognition of face quality-oriented education wisdom evaluation method characterized by comprising
To Face datection, and to facial modeling；
Facial image is pre-processed, face standard drawing is generated；
Face standard drawing is generated into the human face recognition model based on deep learning, calculates and obtains the face based on human face recognition model
Cosine similarity in feature vector and database between existing face database vector；
Result judgement face recognition result based on cosine similarity；
If recognition of face success enters overall evaluation system and carries out overall merit, and generates overall merit handbook.
2. recognition of face quality-oriented education wisdom evaluation system according to claim 1, which is characterized in that described to be examined to face
It surveys, and to facial modeling, comprising: Face datection is carried out based on LBP feature and AdaBoost, and is based on points distribution models
Algorithm positions 68 human face characteristic points；
It is described that Face datection is carried out based on LBP feature and AdaBoost, including cromogram is converted into grayscale image, in any neighborhood
It is interior, using centre of neighbourhood pixel as threshold value, the gray value of 8 adjacent pixels is compared with it, if neighborhood territory pixel value is greater than
Center pixel value, then the position of the pixel is marked as 1, is otherwise 0；
8 pixels in neighborhood, which are compared, generates 8 bits, and 8 bit is converted to decimal number, is calculated
Obtain the LBP value of the centre of neighbourhood pixel.
3. recognition of face quality-oriented education wisdom evaluation system according to claim 2, which is characterized in that described to be calculated
The LBP value of the centre of neighbourhood pixel, comprising:
If (xc,yc) be center pixel coordinate, p be neighborhood p-th of pixel, ipFor the gray value of neighborhood territory pixel, icCentered on
The gray value of pixel, LBP (xc,yc) be center pixel LBP value；X is neighborhood territory pixel value-center pixel value difference, and s (x) is
Sign function；
Then
Further, the LBP value of the centre of neighbourhood pixel is calculated,
4. recognition of face quality-oriented education wisdom evaluation system according to claim 3, which is characterized in that described to be based on LBP
Feature and AdaBoost carry out Face datection, further includes: by the LBP (xc,yc) it is sent into AdaBoost classifier, and divided
Class.
5. recognition of face quality-oriented education wisdom evaluation system according to claim 4, which is characterized in that the AdaBoost
Classifier includes multiple cascade classifiers；
It is described by the LBP (xc,yc) it is sent into AdaBoost classifier, and classify, comprising: the AdaBoost classifier
It is detected using multiple dimensioned sliding window, window is put into multiple cascade classifiers by the window that each scale interception size is 20*20
Middle judgement is face, if face, then the window is by all cascade classifiers, if not face, then the window
It is excluded in a certain cascade classifier.
6. recognition of face quality-oriented education wisdom evaluation system according to claim 2, which is characterized in that described to face figure
As pretreatment includes: to construct face 3D master pattern, estimation human face posture, progress facial image school using face earth axes
Just, it cuts and is aligned, illumination pretreatment is carried out to facial image, eliminate influence of the illumination to face.
7. recognition of face quality-oriented education wisdom evaluation system according to claim 6, which is characterized in that described to face figure
As pretreatment further include: carry out Glasses detection to facial image, extract glasses image, obtain denoising image.
8. recognition of face quality-oriented education wisdom evaluation system according to claim 7, which is characterized in that described to face figure
As carrying out Glasses detection, glasses image is extractd, obtains denoising image, comprising: if detecting including glasses image, using maximum kind
Between variance method carry out binary conversion treatment, obtain bianry image, the not connected smaller image border in part eliminated using opening operation, so
Isolated marginal point is removed using the method for connected domain afterwards, removal center of gravity is in the edge below of picture altitude 1/2, using 2 times
The method of iteration closed operation connects remaining edge, then transversal scanning image, by edge breaks length lower than 1/6 picture traverse
Edge is attached, and forms rims of spectacle mask image, rims of spectacle positioning is completed, finally according to mask image and glasses periphery
The colour of skin repairs frame using linear interpolation mode, obtains denoising image.
9. recognition of face quality-oriented education wisdom evaluation system according to claim 1, which is characterized in that described by face mark
Quasi- figure generates the human face recognition model based on deep learning, calculates and obtains feature vector and database based on human face recognition model
In cosine similarity between existing face database vector, comprising: by deep learning training human face recognition model, be based on face
Identification model extracts face characteristic to face standard drawing, carries out PCA dimensionality reduction to the face feature vector of extraction, calculates face spy
Levy the distance between the human face characteristic point that vector represents.
10. -9 described in any item recognition of face quality-oriented education wisdom evaluation systems according to claim 1, which is characterized in that institute
Stating overall evaluation system includes all evaluation units, moon evaluation unit, term evaluation unit and comprehensive score result unit, hair
Unit, personal integral unit, comparation and assessment unit are suggested in exhibition；
The overall merit includes academic record evaluation, morality evaluation, the evaluation of body and mind Qualities Evaluation, affective state, campus
Performance evaluation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811345573.2A CN109377429A (en) | 2018-11-13 | 2018-11-13 | A kind of recognition of face quality-oriented education wisdom evaluation system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811345573.2A CN109377429A (en) | 2018-11-13 | 2018-11-13 | A kind of recognition of face quality-oriented education wisdom evaluation system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109377429A true CN109377429A (en) | 2019-02-22 |
Family
ID=65384773
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811345573.2A Pending CN109377429A (en) | 2018-11-13 | 2018-11-13 | A kind of recognition of face quality-oriented education wisdom evaluation system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109377429A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109993127A (en) * | 2019-04-03 | 2019-07-09 | 浙江口信科技有限公司 | A kind of facial image recognition method based on artificial intelligence |
| CN110443203A (en) * | 2019-08-07 | 2019-11-12 | 中新国际联合研究院 | The face fraud detection system counter sample generating method of network is generated based on confrontation |
| CN111144338A (en) * | 2019-12-30 | 2020-05-12 | 深圳纹通科技有限公司 | Feature matching algorithm based on feature point topological structure |
| CN113642450A (en) * | 2021-08-09 | 2021-11-12 | 深圳市英威诺科技有限公司 | Video face recognition method, system and storage medium |
| CN116109456A (en) * | 2023-04-03 | 2023-05-12 | 成都大学 | Comprehensive evaluation method and system for intelligent education, electronic equipment and storage medium |
| CN116757524A (en) * | 2023-05-08 | 2023-09-15 | 广东保伦电子股份有限公司 | Teacher teaching quality evaluation method and device |
| CN116935478A (en) * | 2023-09-13 | 2023-10-24 | 深圳市格炎科技有限公司 | Emotion recognition method and system for intelligent watch |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020579A (en) * | 2011-09-22 | 2013-04-03 | 上海银晨智能识别科技有限公司 | Face recognition method and system, and removing method and device for glasses frame in face image |
| CN103915000A (en) * | 2014-04-14 | 2014-07-09 | 徐州市广联科技有限公司 | Online examination method for actual operation and training |
| CN106326868A (en) * | 2016-08-26 | 2017-01-11 | 江苏华通晟云科技有限公司 | Face identification method based on cosine similarity measure learning |
| CN106709442A (en) * | 2016-12-19 | 2017-05-24 | 深圳乐行天下科技有限公司 | Human face recognition method |
| CN107316261A (en) * | 2017-07-10 | 2017-11-03 | 湖北科技学院 | A kind of Evaluation System for Teaching Quality based on human face analysis |
| CN107331247A (en) * | 2017-07-20 | 2017-11-07 | 武汉依迅北斗空间技术有限公司 | Wire examination method and device are trained in a kind of driving training |
-
2018
- 2018-11-13 CN CN201811345573.2A patent/CN109377429A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020579A (en) * | 2011-09-22 | 2013-04-03 | 上海银晨智能识别科技有限公司 | Face recognition method and system, and removing method and device for glasses frame in face image |
| CN103915000A (en) * | 2014-04-14 | 2014-07-09 | 徐州市广联科技有限公司 | Online examination method for actual operation and training |
| CN106326868A (en) * | 2016-08-26 | 2017-01-11 | 江苏华通晟云科技有限公司 | Face identification method based on cosine similarity measure learning |
| CN106709442A (en) * | 2016-12-19 | 2017-05-24 | 深圳乐行天下科技有限公司 | Human face recognition method |
| CN107316261A (en) * | 2017-07-10 | 2017-11-03 | 湖北科技学院 | A kind of Evaluation System for Teaching Quality based on human face analysis |
| CN107331247A (en) * | 2017-07-20 | 2017-11-07 | 武汉依迅北斗空间技术有限公司 | Wire examination method and device are trained in a kind of driving training |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109993127A (en) * | 2019-04-03 | 2019-07-09 | 浙江口信科技有限公司 | A kind of facial image recognition method based on artificial intelligence |
| CN110443203A (en) * | 2019-08-07 | 2019-11-12 | 中新国际联合研究院 | The face fraud detection system counter sample generating method of network is generated based on confrontation |
| CN110443203B (en) * | 2019-08-07 | 2021-10-15 | 中新国际联合研究院 | Confrontation sample generation method of face spoofing detection system based on confrontation generation network |
| CN111144338A (en) * | 2019-12-30 | 2020-05-12 | 深圳纹通科技有限公司 | Feature matching algorithm based on feature point topological structure |
| CN111144338B (en) * | 2019-12-30 | 2022-03-22 | 深圳纹通科技有限公司 | Feature matching algorithm based on feature point topological structure |
| CN113642450A (en) * | 2021-08-09 | 2021-11-12 | 深圳市英威诺科技有限公司 | Video face recognition method, system and storage medium |
| CN116109456A (en) * | 2023-04-03 | 2023-05-12 | 成都大学 | Comprehensive evaluation method and system for intelligent education, electronic equipment and storage medium |
| CN116109456B (en) * | 2023-04-03 | 2023-07-28 | 成都大学 | Comprehensive evaluation method and system for intelligent education, electronic equipment and storage medium |
| CN116757524A (en) * | 2023-05-08 | 2023-09-15 | 广东保伦电子股份有限公司 | Teacher teaching quality evaluation method and device |
| CN116757524B (en) * | 2023-05-08 | 2024-02-06 | 广东保伦电子股份有限公司 | Teacher teaching quality evaluation method and device |
| CN116935478A (en) * | 2023-09-13 | 2023-10-24 | 深圳市格炎科技有限公司 | Emotion recognition method and system for intelligent watch |
| CN116935478B (en) * | 2023-09-13 | 2023-12-22 | 深圳市格炎科技有限公司 | Emotion recognition method and system for intelligent watch |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109377429A (en) | A kind of recognition of face quality-oriented education wisdom evaluation system | |
| CN107194341B (en) | Face recognition method and system based on fusion of Maxout multi-convolution neural network | |
| CN106295522B (en) | A kind of two-stage anti-fraud detection method based on multi-orientation Face and environmental information | |
| CN106878677B (en) | Student classroom mastery degree evaluation system and method based on multiple sensors | |
| CN106372581B (en) | Method for constructing and training face recognition feature extraction network | |
| US11531876B2 (en) | Deep learning for characterizing unseen categories | |
| CN103902961B (en) | Face recognition method and device | |
| CN107169455B (en) | Face attribute recognition method based on depth local features | |
| CN108229330A (en) | Face fusion recognition methods and device, electronic equipment and storage medium | |
| CN107967458A (en) | A kind of face identification method | |
| CN109033938A (en) | A kind of face identification method based on ga s safety degree Fusion Features | |
| Sahbi et al. | A Hierarchy of Support Vector Machines for Pattern Detection. | |
| CN109255289B (en) | Cross-aging face recognition method based on unified generation model | |
| CN106778496A (en) | Biopsy method and device | |
| CN109815826A (en) | The generation method and device of face character model | |
| CN108182409A (en) | Biopsy method, device, equipment and storage medium | |
| CN110503000B (en) | Teaching head-up rate measuring method based on face recognition technology | |
| CN112784763A (en) | Expression recognition method and system based on local and overall feature adaptive fusion | |
| CN109685713B (en) | Cosmetic simulation control method, device, computer equipment and storage medium | |
| CN104517097A (en) | Kinect-based moving human body posture recognition method | |
| CN105205449A (en) | Sign language recognition method based on deep learning | |
| CN107220598B (en) | Iris image classification method based on deep learning features and Fisher Vector coding model | |
| CN110175501A (en) | More people's scene focus recognition methods based on recognition of face | |
| CN106599785A (en) | Method and device for building human body 3D feature identity information database | |
| CN104063721A (en) | Human behavior recognition method based on automatic semantic feature study and screening |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190222 |
|
| RJ01 | Rejection of invention patent application after publication |